Bayesian models and algorithms for protein beta-sheet prediction

This is the latest version of this item.

Official URL: http://dx.doi.org/10.1109/TCBB.2008.140

Prediction of the three-dimensional structure greatly benefits from the information related to secondary structure, solvent accessibility, and non-local contacts that stabilize a protein's structure. Prediction of such components is vital to our understanding of the structure and function of a protein. In this paper, we address the problem of beta-sheet prediction. We introduce a Bayesian approach for proteins with six or less beta-strands, in which we model the conformational features in a probabilistic framework. To select the optimum architecture, we analyze the space of possible conformations by efficient heuristics. Furthermore, we employ an algorithm that finds the optimum pairwise alignment between beta-strands using dynamic programming. Allowing any number of gaps in an alignment enables us to model beta-bulges more effectively. Though our main focus is proteins with six or less beta-strands, we are also able to perform predictions for proteins with more than six beta-strands by combining the predictions of BetaPro with the gapped alignment algorithm. We evaluated the accuracy of our method and BetaPro. We performed a 10-fold cross validation experiment on the BetaSheet916 set and we obtained significant improvements in the prediction accuracy.